do user fees exclude low-income people from … user fees exclude low-income people from...

Journal of Leisure Research Copyright 20002000, Vol. 32, No. 3, pp. 341-357 National Recreation and Park Association

Do User Fees Exclude Low-income People fromResource-based Recreation?

Thomas MoreUSDA Forest Service, Northeast Research Station

Thomas StevensDepartment of Resource Economics, University of Massachusetts

A mail survey of New Hampshire and Vermont households shows that althoughuser fees are widely accepted, they may substantially reduce participation inresource-based recreation by those earning less than $30,000 per year. For ex-ample, 23% of low-income respondents indicated that they had either reduceduse or gone elsewhere as a result of recent fee increases, while only 11% ofhigh-income users had made such changes. A conjoint analysis also suggeststhat low-income respondents are much more responsive to access fees thanhigh-income respondents. And we find mat a $5 daily fee for use of publiclands would affect about 49% of low-income people as compared to 33% ofhigh-income respondents. We conclude that potential impacts of this magni-tude highlight several critical problems in the design of recreation fee pro-grams.

KEYWORDS: Recreation fees, economics, low-income users, public policy

Introduction

Congressional authorization of the current fee demonstration program,which permits public agencies to charge for access to federal lands, has re-invigorated the debate over recreation fees. Although many arguments bothfor and against fees have been advanced (cf., Harris & Driver 1987; Cromp-ton & Lamb, 1986; Shultz, McAvoy, & Dustin, 1988), few are as compellingor as central to the debate as the idea that fees may exclude low-incomeusers from access to public recreation areas. Those opposed to fees pointout that exclusionary pricing raises fundamental questions about the socialpurposes of public recreation (More, 1999). Those who favor fees admit thatexclusionary pricing might be an issue in urban areas, but argue that it isnot important in resource-based recreation because: 1) Low-income peopleare already priced out by high travel and equipment costs (Clawson &Knetsch, 1966; Vaux, 1975), and 2) Everyone, including low-income people,must make choices about how to spend their money, so it is hardly surprisingthat resource-based recreation ranks relatively low among the prioritiesof low-income people (a version of the economic efficiency argument [seeRosenthal, Loomis & Peterson, 1984]).

The authors thank Catherine Howard, Pat Sullivan, Peggy Cialek, Rebecca Oreskes, ThomasMoore, and Greg Wright for their assistance in different phases of this project. Thomas Morecan be contacted at USDA Forest Service, PO Box 968, Burlington, VT 05402.

341

342 MORE AND STEVENS

To appreciate the significance of these arguments, it is important tospecify the social context within which the fee programs have been initiated.Although the United States has always considered itself to be a middle-classcountry, the actual historical record reveals a different picture. While incomeinequality has always been with us, it rose rapidly after the Civil War, peakingin the 1920's (Hurst, 1998). During the 1930's, however, social programsdesigned to relieve the effects of the Great Depression and the subsequenteconomic expansion associated with World War II combined to produce aprosperity in the 1950's and 1960's that was so widely shared that it wasdubbed the "Golden Era" of the U.S. economy (Cassidy, 1995). The actualyear of greatest income equality proved to be 1968 (U.S. Bureau of the Cen-sus, 1996). This prosperity began to fade during the 1970's, however, andthe period between 1973 and 1993 witnessed massive increases in incomeinequality. During this time period, income levels for the bottom 40% ofAmerican families declined in real terms, while the costs of rent, utilities,and food rose with inflation. By contrast, the incomes of the top 20% grewso rapidly that, in 1993, they received 48.2% of the country's aggregate in-come (Mishel & Bernstein, 1993; Cassidy, 1995).

Income inequality has continued to grow throughout the 1990's (U.S.Bureau of the Census, 2000a). Following a major recession early in the dec-ade, the American economy has emerged into what may be its longest-running financial boom. For example, since 1995, the stock market has cre-ated more than $5 trillion in new financial wealth (Frank, 1999). Yet over50% of all Americans own no stock at all ("Economist urges," 1998) andhave not participated in these gains. In fact, most of the gains have gone tolarge stockholders so that economist Edward Wolff (cited in Cassidy, 1999)estimates that 85% of the country's financial wealth is now controlled by thetop 10% of the population. By contrast, in 1998 the median household in-come in the U.S. had risen to $38,885, the family income for married-couplefamilies was $54,276, and for female householders with no husband presentthe median income was $24,393 (U.S. Bureau of the Census, 2000a). Overall,the simple fact that half the households in the U.S. have annual incomesbelow $38,885 indicates the extent of income disparity, and some economistsbelieve that it would take 20 years of sustained economic growth to returnus to the prosperity of the late 1960's (Marger, 1999).

The growing stress on family incomes over the past 30 years also hasimpacted political discourse. Many people blamed their declining incomeson big government (and government spending) so steps were taken to re-duce public-sector spending. For recreation agencies, declining budgets oc-curred in the face of burgeoning demand, placing huge pressure on theagencies and making it difficult to sustain recreation opportunities (Morton,1997), especially given a multibillion dollar backlog of deferred maintenance(General Accounting Office, 1998). Little wonder, then, that agency admin-istrators eagerly turned to fees as a way out and financial self-sufficiencybecame a dominant park philosophy (LaPage, 1994; Leal & Fretwell, 1997).Yet as fee programs were initiated, public appropriations often declined ne-

DO USER FEES EXCLUDE? 343

cessitating further fee increases. While many visitors have been relatively un-affected by these increases, questions remain about the impact of fees onspecific subgroups such as low-income people (General Accounting Office,1998).

These, then, are the major players in the current fee debate: legislatorsanxious to shrink the size of government or to limit spending increases,agencies eager to supplement their budgets through fee programs, and apublic that has decided differences in their ability to pay, coupled with acurrent ambivalence about taxes and government programs. And it is relativeto the varying goals of these groups that we must evaluate fee programs byasking the questions implied in our title: Will fees price low-income usersout of state and national parks and forests? How do nonusers feel about feesversus taxes? What alternatives to fee programs exist? In this paper we ex-amine the attitudes of New Hampshire and Vermont residents regardingthese issues, focusing explicitly on the relationships between these attitudesand income.

Background and Methods

There can be little doubt that price increases reduce use rates, partic-ularly where demand is elastic. For example, Schroeder and Louviere (1999)found that entry fees for Chicago area sites had a significant impact on sitechoice, with choice probability declining steadily over the entire fee range.A $3 fee decreased the site choice probability for their entire sample bynearly 40%. Similarly, Richter and Christensen (1999), using data from asurvey of Desolation Wilderness (CA) visitors, found general visitation de-clined with price increases. Where demand is relatively inelastic, fee increasesare less likely to impact total visitation. Thus the U.S. National Park Serviceexpected that overall visitation would not be affected, but the sites most likelyto experience decreased demand were lesser known sites and sites usedmostly by the surrounding communities (cited in Schroeder & Louviere,1999).

While total visitation is certainly an issue, our research is primarily con-cerned with the impact of fees on different income groups. Here, there ismuch less information. In their review of the Demonstration Program, theGeneral Accounting Office (GAO) concluded that fees have had little impacton overall visitation and that visitor surveys show that they have been gen-erally well received, but noted that "some groups have expressed concernsabout gaps in the research. For example, many completed visitor surveys donot address the impact of fees on some types of visitors, such as those withlow incomes" (GAO 1998, p. 3). Therefore, the GAO report called for moreresearch on the impact of fees on low-income people.

In one study that examined this issue specifically, Reiling, Chung andTrott (1992) found that increasing camping fees at Maine state parks wouldhave a larger effect on low-income campers as compared to high-incomecampers. Although low-income campers in Maine camped more than upper-


income campers when fees were low, they dropped out of the market quicklyas fees rose. In another study, Reiling et al. (1994) found that low-incomeusers of day-use Army Corps of Engineers recreation facilities were moresensitive to fees than upper-income users and consequently high fees woulddisplace a greater proportion of low-income users. The displacement wasconfirmed by Schneider and Budruk (1999) who found that about one-thirdof the visitors they surveyed at a southwestern national forest beach area hadaltered their visitation in response to a fee program. Of these, 62% visitedthe area less often, and over 50% chose to visit free sites within the forest.

A significant difficulty with many existing fee studies is their reliance onvisitor surveys (as opposed to surveys of the general population). Many on-site visitor surveys have found the majority of users favor fees (Leuschner etal., 1987; Williams et al., 1999) particularly when the fees collected are usedfor improvements at the collection site. For example, in 1997 only 17% ofvisitors sampled at National Park Service sites said that fees were too high,and 12% actually claimed that fees were too low (Farmer, 1998). Similarly,Williams et al., (1999) found that most users of California's Desolation Wil-derness considered fees to be acceptable, although they rated wilderness feesas less acceptable than fees for other forms of forest recreation. As notedabove, such studies often argue that low-income people have already beenpriced out of the market for forest recreation due to high equipment andtravel costs (Clawson & Knetsch, 1966; Vaux, 1975). But such reasoning maybe overly general. The American working class represents a broad spectrumof the population which is neither poor nor immobile but which may beunderrepresented in visitor surveys due to existing fees (More, 1999). Sup-pose, for example, that you owned a movie theater that charged $35 perticket. A survey of the few people who came might well reveal that they weresatisfied and supported the fee. What you would miss would be the opinionsof those who never showed up because of the fee. In the private sector, ofcourse, you would be unlikely to charge $35 per ticket: competition wouldkeep prices in line, and reference prices—what people are used to paying—would operate similarly (McCarville & Crompton, 1987; McCarville, 1996).In the public sector, however, there generally is no competition to regulateprices and many experiences are unique. For example, a float trip down theColorado River through the Grand Canyon (a unique experience for whichthere are not likely to be many reference prices) now costs $100 per personto get on the waiting list and $100 per person to access the river (Hanscom,2000). And there are even some user groups that support fees simply becausethey will exclude other users (Winter et al., 1999).

The above considerations point to the necessity of basing public-sectorpricing policy on studies of the general population rather than solely relyingon on-site "customer" surveys. In theory, general population surveys can cap-ture responses from a broad social and economic spectrum. In practice, un-fortunately, obtaining responses from low-income people can be problem-atic; national forest system planners, for example, report that they often havegreat difficulty involving low-income people in their planning efforts. To


overcome these problems, we conducted a general population mail surveywhich utilized an unusual stratified sampling technique designed to boostresponses by low-income people. We began by dividing both New Hampshireand Vermont into quadrants of nearly equal size. For each quadrant, 1990census data were used to select the four communities with the lowest medianhousehold income. The communities ranged in size from small, rural townsto small cities of nearly 30,000 residents. A random sample of residents ineach of these 32 communities was then selected, resulting in an overall sam-ple of 1,000 New Hampshire and Vermont residents. This procedure involvedmaking a methodological tradeoff. With it we were able to increase returnsfrom low-income people, but at the expense of being able to make statewideprojections to the population an a whole. While the study results could, intheory, be generalized to the population of residents of low-income NewHampshire and Vermont communities, it may be best to treat the compari-sons made in this study as sample specific, i.e., as a randomly-selected sampleof low-, middle-, and upper-income people.

The questionnaire was adapted from Dillman's (1978) total designmethod, and the survey was mailed in January, 1999. There were 138 unus-able addresses and 296 respondents (161 from New Hampshire and 135 fromVermont) giving an adjusted response rate of 34%, about average for aca-demic mail surveys of the general population (Mitchell & Carson, 1989). Wedid not formally re-contact nonrespondents as the sample design precludedmaking statewide population projections.

Seventy-one percent of respondents were male. The average respondentwas 54 years old, had about 14 years of education, and a household incomeof between $30,000 and $45,000 per year. About 28% of the respondentsreported an annual household income of less than $30,000 per year, 54%had incomes between $30,000 and $74,999 per year, and 18% reported in-comes exceeding $75,000 per year.

Unfortunately, directly comparable socioeconomic and demographicdata for the population of low-income communities in both states are notavailable, which limits our ability to draw general inferences. However, themedian household income for 1998 was $34,592 in Vermont and $40,854 inNew Hampshire. And, in 1998, about 31% of the adult (over age 24) resi-dents of New Hampshire and Vermont were between 45 and 64 years old ascompared to 41% of our respondents. Based on these figures, we believeour sample, although close to being representative on some dimensions, wasstill weighted slightly toward the more educated and affluent, and signifi-cantly toward males. Again, it is probably best to treat these results as rep-resenting randomly selected low-, middle-, and upper-income people.

For analysis, we divided respondents into three income groups: low (lessthan $30,000 per year), middle ($30,000 to $74,999 per year), and high(greater than $75,000 per year). On average, low-income respondents wereolder, had less education, and were more likely to be female as comparedwith middle- and upper-income respondents (Table 1). These differences,which were statistically significant (p < 0.05), are obviously important when


TABLE 1Characteristics of Income Groups

Characteristics Low Income Middle Income High Income

NAverage ageGender (%male)Average education (years)

6958.66212.79

13351.47514.60

4551.38416.78

analyzing and interpreting the survey results. Consequently, we used two teststo determine statistical significance: The standard "z" test was used to testfor differences in population proportions, while regression analysis was usedto isolate the effects of income, if any, on respondent's behavior or attitudeswhile holding age, education, and gender constant.

Results

To determine the impact of user fees on the three income groups, eachrespondent was asked multiple questions. First, respondents were asked howoften they participated in various outdoor activities during the summer andfall of 1998. Compared to the highest income group, low-income respon-dents participated more frequently in fishing, backpacking, hunting and tripsto watch birds and wildlife (Table 2). However, according to the standardtest for differences in population proportion, reported participation ratesfor these activities did not differ statistically by income category. Moreover,when participation rates were regressed against respondent's age, education,gender and income group, income differences were not a statistically signif-icant factor. Hunting trips, for example, showed a statistically significant de-

TABLE 2Reported Participation Rates by Activity and Income Group, Summer and Fall,

1998*

Activity

PicnickingFishingCampingBackpackingMountain bikingHuntingTrips to watch birds

and wildlife

Low Income

3.9210.812.84

.821.447.416.52

Middle Income

4.756.171.29

.353.275.817.20

High Income

3.937.772.93

.171.345.483.35

"Rate expressed as average number of times participating.


crease with age, education and female gender, but income group was not astatistically significant factor (p < 0.05) in explaining the variation in hunt-ing trips per respondent.

Next, we asked if entrance, parking or access fees were a major factorinfluencing decisions about participation in these activities. We also askedabout areas visited, frequency of visits and fees paid between May 25 andDecember 31, 1998. For those in the low- and middle-income groups, en-trance, access or parking fees were more likely to be a major factor in activityparticipation decisions. Only about 11 % of the upper-income group said thatfees were a major factor in participation decisions, while 16.2% and 18.2%of the middle- and low-income respondents, respectively, said that fees werea major factor. These differences were not statistically significant, however.

Low-income respondents were generally less likely to have visited stateparks, state forests, or national forests in either New Hampshire or Vermontduring the summer, 1998. Only 34.4% of the lowest income group visitedone of these areas, while 51.2% and 53.2% of middle- and upper-incomeresidents visited these areas, respectively. The difference in visitation betweenthe low- and upper-income groups was statistically significant (p < 0.05)when considered alone. However, when the effects of gender, age and edu-cation were considered, income was not a statistically significant factor ex-plaining this difference.

We also questioned respondents about their plans for participation inwinter and spring activities for 1999. As expected, the lowest income grouphad lower planned participation rates in several of the activities listed (Table3) but most differences were not significant. Planned participation rates var-ied dramatically by income class for both cross-country and downhill skiing,however. And regression analysis showed that the differences between thelow- and high-income groups were statistically significant (p < 0.05) fordownhill but not for cross-country skiing, holding age, gender and educationconstant.

When asked if an increase in access fees of $5 per visit would influenceparticipation in any of these winter/spring activities, 49.2%, 36.7%, and33.3% of the low-, middle-, and upper-income groups said yes, respectively.

TABLE 3Planned Participation by Activity and Income Group, Winter and Spring, 1999b

Activity Low Income Middle Income

3.161.561.221.611.44

bAverage number of times planned.

X-Country skiingDownhill skiingIce fishingSnowmobilingSnow shoeing

1.52.50

1.50.48.81

High

34

1

Income

.26

.25

.35

.54

.10


A logit regression analysis (with the dependent variable denned as 1 if theindividual said they would be affected by an increase in fees and 0 otherwise)showed that these differences are statistically significant (p < 0.05); olderrespondents and male respondents were less likely to be affected, while low-income respondents were more likely to be impacted, all else held constant.

Respondents then were asked a series of questions about their attitudestoward changes in fees for access and related services, and about who shouldpay these fees. When asked if they would favor a policy that maintains presentservices at parks and forests by increasing fees or a policy that keeps fees atpresent levels but reduces services, 67% of the mid- and upper-income re-spondents preferred to maintain services by increasing fees, while 57% ofthe low-income respondents chose this option. This difference was not sta-tistically significant, however.

The services that respondents felt they could most do without were trailsfor ATV's, bikes, etc. (58%); late season campground operation (37%); Sun-day operation of information areas (27%); educational programs (25%); andsome boat access (22%). The services that respondents were most reluctantto give up were restroom maintenance, garbage pickup at both roadsideareas and campgrounds, and law enforcement (Table 4).

We then asked respondents to assume that services at state and federalsites in New Hampshire and Vermont would not be reduced, but that moremoney would have to be raised to cover costs. Respondents were asked howthis fund raising should be accomplished. The upper-income group clearlyfavored increasing access fees, while the lowest income class was much morelikely to favor increased reliance on voluntary contributions or state andfederal taxes (Table 5). Except for the sales tax option (see Table 5), differ-ences between the lowest and highest income groups with respect to pref-erences about fundraising policies were statistically significant (p < 0.05),

TABLE 4Desirability of Service Reductions

Service Reduction Option Percent Approving Reduction

Close some trails to ATV's, bikes, etc. 58.0Early closure of campgrounds 37.4Close information areas on Sunday 27.3Eliminate education efforts 24.9Close some boating access areas 22.4Reduce trail maintenance 17.0Reduce fish / wildlife management 15.7Reduce road maintenance 13.9Reduce law enforcement 11.5Eliminate garbage pickup at campgrounds 9.1Eliminate garbage pickup at roadside areas 8.0Reduce restroom maintenance 4.6


TABLE 5Preferences of How More Money Should be Raised to Cover Cost, by Income Groupc

Option

Increase parking or access feesIncrease state or federal taxesIncrease sales tax on outdoor equipmentIncrease reliance on voluntary contributions

(labor or $ or both)Other

Low Income

26.216.918.533.8

4.6

Middle Income

38.915.914.324.6

6.3

High Income

60.54.7

18.614.0

2.3

'Percent of respondents selecting option.

while holding effects due to respondents' age, education and gender con-stant.

When asked who should pay the most to maintain and improve parksand forests in New Hampshire and Vermont, 43 percent of the low-incomegroup said "all taxpayers," while only 27% of the upper-income group gavethis response (a statistically significant difference, p < 0.05); 45% of upper-income respondents favored payment by campers or consumptive users(hunters, anglers). Seventy-three percent of low-income respondents eitheragreed or strongly agreed with the statement that "state parks and nationalforests should be available to everyone regardless of their ability to pay."Sixty-four percent of high-income respondents agreed with this statement,but this difference was not statistically significant (p > 0.05).

One potential effect of fee increases is that working-class families mightreduce or eliminate visits to state or federal parks and forests. When asked,a majority of low-income respondents (53%) felt that this was an importantor very important policy consideration; however, 58% of the highest-incomerespondents expressed the same sentiment, suggesting a broad-based rec-ognition of the social importance of these services.

We also asked an open-ended question: "Entrance and access or parkingfees to most public outdoor recreation areas in New Hampshire and Vermonthave increased over the past five years; how have you responded to thischange?" Although most people reported that they had been unaffected bythese increases, there were important differences across income categories.Eighty-four percent of the high-income respondents, as compared to 60% ofthe low-income respondents, said that fee increases had not affected themor that they had "just paid" the increases. More importantly, 23% of the low-income respondents indicated that they had either reduced their use or goneelsewhere, while only 11% of high-income users had made such changes.Both of these differences were statistically significant (p < 0.05).

Finally, we presented respondents with four scenarios for use in a con-joint analysis. Option 1 was to stay home, option 2 was to visit a Vermontstate park, option 3 was to visit the Green Mountain National Forest (VT),


and option 4 was to visit the White Mountain National Forest (NH). Foroptions 2, 3, and 4 there were two levels of garbage pickup (none, full), twotypes of toilet facilities (pit, flush), three levels of increase in wildlife popu-lation (0, 10% and 25%), and three alternative fee levels per visit ($1, $2,$5). These levels were randomly assigned to the options in each survey. Re-spondents rated each option on a scale of 1 to 5, where 5 represents theoption, if any, respondents would definitely choose.

In a conjoint analysis, respondent ratings are assumed to be proxies forindividual utility. Suppose, for example, that individual utility associated withforest-based outdoor recreation is expressed by:

u'(p\ q\ m, z) (1)

where p' is the fee associated with recreation option i, q' is a vector of otherattributes associated with option i (wildlife, garbage pickup, type of toiletfacility), m is the respondent's income and z is a vector of individual char-acteristics, like age.

Assuming that utility is related to individual ratings via a transformationfunction, then:

r{(p\ q\ m, z) = &[u\ (ft, q\ m, z)] (2)

where i' is the rating.Since we assume that individual ratings depend on option attributes and

the respondent's socioeconomic characteristics, the following empirical re-lationship was estimated:

Rating = a0 + (31 garbage pickup + (32 type of toilet facility + (33 increasein wildlife population + |34 fee + (35 respondent's age + @6 dummyvariable for Green Mountain National Forest + (37 dummy variablefor White Mountain National Forest + (3g dummy variable for NewHampshire resident. (3)

where a0 and (^ through |38 are estimated coefficients. Separate models(equation 3) were estimated for each of the three income classes. The resultsreported in Table 6 (for the low- and high-income classes) indicate that forboth groups, ratings increased with garbage pickup, flush toilets, the GreenMountain National Forest option and the White Mountain National Forestoption. Ratings decreased with age of the respondent, access fee and resi-dence in New Hampshire.

Of particular importance is that the low-income group was much moreresponsive to access fees as compared to the high-income group. That is, theaccess fee, which was not a statistically significant factor for the high incomegroup was statistically significant [p < 0.01) for the low-income group.1

'The reported results were derived from the OLS estimating technique. Since the range of thedependent variable is limited, an ordered logit model was also estimated. The results of theordered logit model, available from the senior author, were very similar.


TABLE 6Regression Results of Rating Model

Low-Income Group High-Income GroupVariable Estimated Coefficient"1 Estimated Coefficient11

Intercept 4.682 (8.23)*** 3.934 (6.43)***Garbage pickup (1 if full, 0 otherwise) .749 (3.20)*** .057 (.213)Type of toilet (1 if flush, 0 otherwise) .534(2.28)** .419(1.58)Increase in wildlife population (%) -.009 (.80) .019 (1.48)Access fee ($) -.229(3.27)*** -.142(1.74)*Respondent's age (years) -.022(2.88)*** -.018(1.85)*Green Mountain (1 if Green Mountain option, .045 (.16) .028 (0.9)

0 otherwise)White Mountain (1 if White Mountain option, .488 (1.74)* .792 (2.48)**

0 otherwise)New Hampshire resident (1 if resident, 0 -.285 (1.20) -.432 (1.59)

otherwise)# 2 = .15 fi2 = .14

"•Absolute t values in parentheses.***The difference is significant, p < 0.01**The difference is significant, p < 0.05*The difference is significant, p < 0.10

It is also important to remember however, that the choices presentedrepresent hypothetical situations; actual behavior may be different. More-over, the attributes associated with each option do not represent many of thecomplexities associated with actual decisions. On the other hand, this typeof hypothetical analysis provides information about respondent's behaviorthat cannot be quantified in any other way.

Discussion

Two major points arise from these results. First, there is clearly broad-based attitudinal support for fees. Forty percent of the entire sample pre-ferred fees to other methods of raising funds (Table 5), and even low-incomerespondents tended to prefer fees to reductions in services.

Second, it is quite clear that fees have a major discriminatory impact onlow-income people. One multiple measures and across multiple questions,the effects of income are clear, consistent and significant. For example, alarger percentage of low-income people have altered their behavior becauseof fees; low-income people were more likely to prefer fundraising via dona-tions as compared to fees, and low-income people were far more likely tobelieve that all taxpayers should be responsible for financing public lands.

These findings clearly highlight the notion that fees strike low-incomepeople harder than upper-income people. Some high-income respondents


also indicated that they too would be, or have been, displaced—ranging from11% who indicated that increasing fees had changed their participation inthe past five years, to 33% who claimed that they would make changes intheir planned, future participation. This result compares favorably with thoseof Schneider and Budruk (1999) cited above. However, participation changesmay be easier for upper-income people than for low-income people. Rosen-thai, Loomis and Peterson (1984) argue that fees promote efficient resourceallocation by sorting out high-value users (those who are willing to pay at aparticular level) from low-value users (those who are not). Yet the efficiencyapproach, which assumes that all value is captured in a willingness-to-paymeasure, is not necessarily fair since it fails to account for differences in theability to pay. There is less flexibility when budgets are stretched tight, and$5 looms much larger for families making less than $30,000 per year than itdoes for families who earn more than $75,000, all else held constant.

So fees have both broad-based attitudinal support (including supportfrom low-income respondents) and significant exclusionary impact. Arethese findings contradictory? We think not, primarily because they have dif-ferent origins. To us, it seems likely that the attitudinal sentiment arises fromthe dominant line of the political discourse over the past quarter century:no new taxes, reduced government spending. The exclusionary impact, onthe other hand, arises from the distribution of income, a deeply structuralvariable and one which is likely to be a causal factor in the attitudinal re-sponse. That is to say, as the income gap widens, we are likely to see a changein the political discourse; the preferences for funding public parks and rec-reation expressed in this study are likely to be far more malleable than peo-ple's incomes, and there are already indications that the political wind onthese subjects is changing (cf., Frank, 1999).

A second potential explanation concerns a lack of specificity in our ques-tionnaire. Many of our questions about fees and taxes were not specific aboutthe amount of either. It may be that people are quite used to thinking aboutrecreation fees that range from $2 to $5 (and several questions did refer toa $5 fee), while the notion of a tax increase might have sounded vague andsomewhat more threatening. Future research comparing these alternativesshould probably be specific about both.

Since fees do have a significant negative impact on participation by low-income people, how should public agencies respond? Of the usual justifi-cations given for fees, many are little more than attempts to rationalize ex-cluded users, avoiding any moral issues involved. For example, manymanagers focus on agency welfare, turning excluded users into little morethan an accounting problem. Similarly, a focus on resource protection oreconomic efficiency can support fee programs with little consideration ofwhich visitors get excluded. The currently popular "customer" orientationcan accomplish a similar result. Since low-income people are less likely toparticipate in many forms of resource-based recreation, they can simply bedefined as "not our customers." Each of these strategies is in full play inrecreation management and research; what is missing is a sense of publicneed or mission (More, 2000).


On the other hand, many in both the recreation management and re-search communities do recognize the problem and have made sincere at-tempts to mitigate the negative impacts of fees. Crompton and Lamb (1986),for example, recommended charging an appropriate price for a service andthen finding ways, such as free days or rebates, to encourage participationby the needy. And some programs offer admission in exchange for voluntarylabor. Unfortunately, such conceptions tend to underestimate how compli-cated life can get for many low-income people. As Rubin (1994) notes, manyworking-class families have both limited leisure time and inflexible schedules.They may, for example, work two jobs or have jobs on different shifts. Chil-dren and the attendant necessity of childcare arrangements complicate mat-ters further. While unemployment may bring significant leisure, many em-ployed working-class or working poor people are unable to take advantageof a "free Tuesday," and agencies are unlikely to offer free access on week-ends when the greatest potential for revenue generation exists. Life at themargin gets complicated.

If fees are exclusionary and simple tactics like free days or rebates areineffective, what alternatives exist? The obvious long-term solution is to es-tablish an active constituency that will speak on behalf of public funding forparks and recreation programs. Most agencies do, in fact, have such constit-uencies, but clearly their effects have been limited. Perhaps more effectiveare constituencies for individual parks and programs—"friends groups." Un-fortunately, such groups often have only localized effects and cannot con-tribute to the solution of a general policy problem.

What remains, then, is service reductions. A majority of our sample,including low-income people, preferred fees to service reductions. From thestandpoint of excluded users, however, service reductions present the onlyviable options for maintaining or reducing current fees. Reductions may bepossible in some of the park services listed in Table 4; state parks in bothNew Hampshire and Vermont have eliminated garbage pickup in many ar-eas, for example. However, we probably have reached an era when agenciesmust begin to consider closing entire areas and/or facilities. There is a nat-ural resistance to such an idea: when areas are closed, nobody gets anythingand any existing infrastructure may be a target for vandalism. Even moreimportant will be an agency's ability to withstand pressure that comesthrough the political system. Clearly, any decision or service reductions orfacility closure needs to be made on a sound, rational basis; we very muchneed data on the productivity of facilities and the contributions of differentservices to recreation quality. Only with such data can we make informed,defensible decisions. The alternative—relying on exclusionary fees to main-tain services—may undermine the social purpose of public recreation facil-ities.

Conclusions

User fees, although widely accepted, significantly discriminate againstlow-income people. Based upon our data, we estimate that a $5 daily fee for


the use of public lands will significantly impact about 49% of low-incomepeople. Fewer members of the middle- and upper-income groups will beimpacted, and their exclusion is more likely to be discretionary.

Before concluding, two caveats need to be expressed about this study.First, as noted, these results should be treated as sample specific; the natureof the sample precluded making statewide projections. The limited evidencewe have suggests our sample was somewhat weighted toward more educated,affluent males. While some of these factors were held constant by the statis-tical analysis, readers still would be well advised to treat them simply as com-parisons of people in different income groups. Second, New Hampshire andVermont are rural states. This means that the respondents in our study mayhave readily available substitutes for public lands. The results might havediffered if the study had been conducted in an urban state that had fewersubstitute sites and where people are more dependent on public lands foraccess to resource-based recreation.

These caveats aside, visitor exclusion highlights a critical problem inrecreation policy—the problem of public purpose (More, 1999). In general,for the public sector to be involved in an activity, there needs to be a publicpurpose for it—some public need that must be fulfilled, some public goalthat must be accomplished. This is certainly true for public lands, and it isthis public purpose that must drive policy. The national parks, for example,were founded with the dual (and often conflicting) purposes of protectingthe resource and promoting public use and enjoyment (Chase, 1987). Thus,the key question with regard to fees becomes: How do fees enhance or de-tract from an agency's ability to fulfill its mission? The enhancements areobvious, particularly in the area of maintenance and possibly in the protec-tion of fragile environments as well. For example, with $460,000 returnedfrom the fee demonstration project during the 1998 season, the Mt. Baker-Snoqualmie National Forest was able to hire 24 trail maintenance workerswho cleared over 700 miles of trails, improved drainage, and helped to main-tain trailhead toilets and bulletin boards (Bates, 1999). Little wonder thenthat a majority of site managers are pleased with the program, especiallywhen attendance remains relatively constant (Krannich et al., 1999).

Such improvements may be dearly bought, however; while total visitationmay have remained constant, there is no indication in such data if the typeof visitor is affected. Our data clearly suggest that fees have had a negativeimpact on participation for nearly half of low-income households and, as theGAO (1998) report notes, it is a bit too soon to begin predicting attendance;while we currently are enjoying a major economic expansion, what will hap-pen when a decline sets in?

A serious, related concern is the effect of fees on minority participation;as Winter et al., (1999) note, there is remarkably little research on the effectsof fees on peoples of color. Minority participation was not an issue in thepresent study because the populations of New Hampshire and Vermont arelargely Caucasian—86% and 98%, respectively, in 1998 (U.S. Census Bureau,Online 2000b)—and most minorities were concentrated in the large cities


which did not appear in our sample. In the past, however, fees have beenused to systematically discriminate against both minorities and low-incomeusers (Caro, 1975),2 and many minorities face other barriers to participationwhich fees will only compound (Phillip 1999). The effects of fees on minorityparticipation in outdoor recreation is clearly an area that needs further re-search.

If low-income people are, in fact, excluded from public parks and rec-reation areas, then serious policy questions are raised about the very purposeof public recreation. In Vermont, for example, there are 100 plus miles ofshoreline along Lake Champlain, most of which is privately owned. Wealthypeople have access to the lake in numerous ways: some own waterfront prop-erty, others belong to private clubs or marinas. The public sector does ownproperty, much of which serves to preserve access to the lake for the publicat large. However, when agencies begin to act like entrepreneurs seekingself-funding through fees, and low-income people are excluded, the publicpurpose—the very reason for public ownership—is defeated.

Why do we have public beaches, hiking trails, campgrounds or teencenters? For these resources to be legitimately in the public sector, they mustbe fulfilling a public need; a clear sense of public mission and public purposeis essential to the formation of sound recreation policy (More, 2000). Un-fortunately, this point is often lost in current discussions. Many see fees onlyin terms of cash flows—dollars taken in versus operating costs. And the rec-reation research literature is often most concerned with the mechanics ofsetting fees—reference prices, degree of acceptance, revenue optimizing andthe like. Ultimately, however, a strong sense of mission and purpose arefundamental to the successful management of public parks and recreation.Our results suggest that fees undercut this mission: they are a major step inthe gentrification of recreation resources. When the parks are reserved forthe comfortably well-off, will they continue to be publicly necessary?

References

Bates, N. (1999). Paying to play: The future of recreation fees. Parks and Recreation, 34(7), 47-52.

Caro, R. (1975). The Power Broker. New York: Harcourt Brace and Company.Cassidy, J. (1995, October 16). Who killed the middle class? The New Yorker, 113-124.CassidyJ. (1999, January 25). No satisfaction. The New Yorker, 88-92.Chase, A. (1987). Playing God in Yellowstone. New York: Harcourt, Brace and Company.Clawson, M., & Knetsch, J. L. (1966). Economics of outdoor recreation. Baltimore: John Hopkins

Press.

2Caro (1975) specifically documents cases where Robert Moses, who held a number of parkrelated positions in both New York City and New York State, used fees to discourage participationby minorities and low-income people in both city and state parks in the first half of the 20thcentury. Moses also controlled park design, swimming pool, temperatures, and employment/staffing practices to achieve these ends.


Crompton,J., & Lamb, C. (1986). Marketing government and social services. New York: John Wiley& Sons.

Dillman, D. (1978). Mail and telephone surveys: The total design method. New York: John Wiley andSons.

Economist urges financial restraint. (1998, December 31). Burlington (VT) Free Press, p. IB.Farmer, T. (1998). Signpost: Ante up. Backpacker 26,(8) 16.Frank, R. (1999). Luxury fever. New York: The Free Press.General Accounting Office. (1998). Recreation fees: demonstration fee program successful but could be

improved (GAO/RCED-99-7). Washington, DC: General Accounting Office.Hanscom, G. (2000, February 14). Grand Canyon National Park, Arizona. High Country News

32(3), 10.Harris, C, & Driver, B. (1987). Recreation user fees: I. pros and cons. Journal of Forestry 85(5),

25-29.Hurst, C. (1998). Social inequality. Boston, MA:Allyn & Bacon.Krannich, R., Eisenhauer, B., Field, D., Pratt, C, & Luloff, A. (1999). Implications of the National

Park Service Recreation Fee Demonstration Program for park operation and management:Perceptions of NPA managers. Journal of Park and Recreation Administration, 17(3), 35-52.

LaPage, W. (1994). Self-funding state parks, the New Hampshire experience. Parks, 4(2), 22-27Leal, D., & Fretwell, H. (1997). Users must pay to save our national parks. Consumers Research

80(8), 16-19, 24.Leuschner, W. A., Cook, P. S., Roggenbuck, J. W., & Oderwald, R. G. (1987). A comparative

analysis for wilderness user fee policy. Journal of Leisure Research, 79101-114.Marger, M. (1999). Social inequality: Patterns and processes. Mountain View, CA: Mayfield Publishing

Co.McCarville, R. E. (1996). The importance of price last paid in developing price expectations for

public leisure services. Journal of Park and Recreation Administration 14(4), 52-64.McCarville, R. E., & Crompton, J. L. (1987). An empirical investigation of the influence of

information on reference prices for public swimming pools. Journal of Leisure Research 19,223-235.

Mishel, L., & Bernstein, J. (1993). The state of working America, 1992-1993. Armonk, NY: M.E.Sharpe.

Mitchell, R. C. & Carson, R. T. (1989). Using surveys to value public goods: The contingent valuemethod. Washington, DC: Resources for the Future. .

More, T. (1999). A functionalist approach to user fees. Journal of Leisure Research 31(3). (in press).More, T. (2000). Reconceiving recreation policy in an era of growing social inequality. In G.

Kyle (Ed.)., Proceedings of the 1999 Northeast Recreation Research Symposium (pp. 415-419). Gen.Tech. Rep. NE-269. Newtown Square, PA: U.S. Department of Agriculture, Forest Service,Northeastern Research Station.

Morton, P. (1997). Sustaining recreation resources on southern Appalachian national forests.Journal of Park and Recreation Administration 14(2), 121-208.

Phillip, S. (1999). Are we welcome?: African American racial acceptance in leisure activities andthe importance given to children's leisure. Journal of Leisure Research 31(4), 385-403.

Reiling, S., Cheng, H., & Trott, C. (1992). Measuring the discriminatory impact associated withhigher recreational fees. Leisure Sciences 14(2), 121-138.

Reiling, S., McCarville, R., & White, C. (1994). Demand and marketing study at Army Corps ofEngineers Day-Use Areas. Vicksburg, MS: Waterways Experiment Station.

Richter, J. & Christensen, N. (1999). Appropriate fees for wilderness day use: Pricing decisionsfor recreation on public land. Journal of Leisure Research 31(3), 269-280.

Rosenthal, D., Loomis, J., & Peterson, G. (1984). Pricing for efficiency and revenue in publicrecreation areas. Journal of Leisure Research 16, 195-208.


Rubin, L. (1994). Families on the fault line: America's working class. New York: Harper Collins.Schneider, I., & Budrick, M. (1999). Displacement as a response to the federal recreation fee

program. Journal of Park and Recreation Administration 17(3), 76-84.Schroeder, H., & Louviere, J. (1999). Stated choice models for predicting the impact of user

fees at public recreation sites. Journal of Leisure Research 31(3), 300-324.Shultz, J., McAvoy, L., & Dustin, D. (1988). What are we in business for? Parks and Recreation

23(1), 52.U. S. Bureau of the Census. (1996). A brief look at postwar U. S. income inequality. (Current pop-

ulation reports P60-191, 4 p.). Washington, DC.U. S. Bureau of the Census. (2000a). Income 1998. Online: http://www.census.gov/hhes/in-

come/income98/in98med.html.U. S. Bureau of the Census. (2000b). 1950 to 1998 annual time series of state population estimates by

race and Hispanic origin. Online: http://www.census.gov/population/estimates/state/srh/srhus98.txt.

Vaux, H. (1975). The distribution of income among wilderness users. Journal of Leisure Research7(1), 29-37.

Williams, D., Vogt, C, & Vitters, J. (1999). Structural equation modeling of users' response towilderness recreation fees. Journal of Leisure Research 31(3), 245-268.

Winter, P., Palucki, L., & Burkhardt, R. (1999). Anticipated responses to a fee program: The keyis trust. Journal of Leisure Research 31(3), 207-226.

do user fees exclude low-income people from … user fees exclude low-income people from...

Documents